[vc_row full_width=”stretch_row” css=”.vc_custom_1531732107238{background-color: #fcc118 !important;}”][vc_column]

[/vc_column][/vc_row][vc_row css=”.vc_custom_1531730959461{border-bottom-width: 1px !important;background-color: #f9fafb !important;border-bottom-color: #eef3f7 !important;border-bottom-style: solid !important;}”][vc_column css=”.vc_custom_1531891416301{margin-bottom: 0px !important;}”][bsfp-cryptocurrency style=”widget-6″ align=”auto” columns=”2″ scheme=”light” coins=”top-x-coins” coins-count=”8″ coins-selected=”” currency=”USD” title=”Cryptocurrencies” show_title=”0″ icon=”” heading_color=”” heading_style=”default” bs-show-desktop=”1″ bs-show-tablet=”1″ bs-show-phone=”1″ css=”.vc_custom_1531730265600{margin-bottom: 0px !important;}” custom-css-class=”” custom-id=””][/vc_column][/vc_row]

Cypherpunk Jameson Lopp sees need for machine learning to improve Bitcoin hashrate estimates

0

Stop scaring users with your bad KYC flows

Bitcoin’s global network hashrate may seem like an objective metric, but researcher and Cypherpunk Jameson Lopp reveals measuring it precisely is deceptively tricky. In a recent Proof of Work (POW) Summit talk in Prague, Lopp described his “hunt for the real Bitcoin hashrate” by evaluating the accuracy of various estimation algorithms.

As Lopp explained, most hashrate estimates derive from blockchain data like difficulty targets and block times. However, he noted the volatility in estimates over shorter timeframes. “If you’re only using the past 10 blocks, the hashrate can appear much higher or lower than it is,” said Lopp.

By aggregating hashrate data reported directly from mining pools, Lopp created a benchmark for testing blockchain-based estimates. He found the commonly used 1,000 block (~1 week) estimate had just a 3.8% average error rate. Lopp then tried blending multiple estimates, optimizing for accuracy. His best algorithm used 10 estimates from 100 to 1,000 blocks, throwing out short-term low estimates, and yielded a 3.5% error rate.

“There is still room for improvement,” said Lopp, suggesting machine learning could further refine estimates. While acknowledging no direct financial incentive, Lopp’s goal was a more accurate standard metric. “I primarily see this as a way of us trying to converge on consensus for viewing these networks,” he explained.

“This seems like the type of problem that should be great for machine learning because what we’re trying to do is we’re trying to find the optimal number of different parameters and variables that we can tweak to hone in on what the most accurate hash rate estimate would be this is just sort of me nerding out.”

Lopp also discussed the implications of the upcoming Bitcoin halving. He expects hashrate to keep rising aggressively until the event, calling it “well known” and “priced in” by miners. Lopp predicted a negligible drop in hashrate post-halving, despite some miners shutting down.

The complete recording of Lopp’s POW Summit talk is available online. His presentation slides and related research can be found on his blog.

Lopp is the co-founder and CTO of Casa, a company focused on providing secure storage solutions for Bitcoin. Before that, he worked as a software engineer at BitGo, a security service for Bitcoin and other digital currencies.

Lopp is also known for his educational efforts in crypto, frequently writing and speaking on various aspects of Bitcoin. He maintains a comprehensive resource list for technical information about Bitcoin, blockchain technology, and their potential impacts on society. His commitment to privacy and individual sovereignty is well-documented and strongly reflected in his work and public statements.

Posted In: , Mining

Leave A Reply

Your email address will not be published.